基于近似熵的交通流序列趋势变化检测

发布时间：2018-01-08 12:36

本文关键词：基于近似熵的交通流序列趋势变化检测　出处：《北京交通大学学报》2014年06期 　论文类型：期刊论文

【摘要】：交通流趋势变化特征分析是交通流预测的基础.为了提取交通流序列随时间推移所呈现出来的宏观变化规律,提出了一种用于检测交通流序列趋势变化的滑动移除近似熵方法.通过对交通流序列趋势规律进行研究,首先将其细分为上升趋势、平稳波动趋势、下降趋势,然后根据不同趋势变化的时间序列复杂程度不同,建立了滑动移除近似熵方法求解其滑动移除近似熵的值,并根据得到的时间序列提取交通流序列趋势变化.最后以北京市四环路某一断面交通流序列为例,用建立的模型对交通流序列趋势变化进行检测,并与滑动t检验方法结果对比.研究结果表明本文提出的方法能够对交通流序列趋势变化进行检测,且检测结果与实际交通流序列趋势变化比较吻合,研究结论可为短时交通流预测建模提供参考依据.
[Abstract]:Traffic flow trends analysis is the foundation of traffic flow forecasting. In order to change the macro traffic flow sequence extraction with time are presented, for detection of traffic flow sliding trend removed sequence of approximate entropy method is proposed. Based on the traffic flow trend sequence are studied firstly, divides it into the rise the trend, steady trend, a downward trend, then according to the different trends of the time series of different complexity, established the approximate entropy method for solving the removal of sliding sliding removed the ApEn, and according to the time series of traffic flow extraction sequence trend changes. Taking Beijing city four loop one section traffic flow sequence as an example and to detect the traffic flow trend sequence using the model, and compared with the sliding t test method results. The results show that the proposed method is able to Traffic flow series trend change is detected, and the detection result is consistent with the trend change of actual traffic flow sequence. The conclusion can provide a reference for short-term traffic flow prediction modeling.

【作者单位】：北京交通大学交通运输学院;
【基金】：国家自然科学基金资助项目(71340020)
【分类号】：U491.112
【正文快照】： 时间序列的趋势是表示时间序列移动方向的一种高层模式,是其随时间变化呈现出某种持续上升、下降或平稳的状态或规律,且不同的趋势变化体现了时间序列不同的动力学结构.目前,比较常用的时间序列趋势变化分析方法主要有滑动t检验[1],Mann-Kendall[2],累积距平法[3]等方法.城市

【共引文献】